Add VCS repo cloning support (git/hg/svn/fossil)

Smart source detection: like Hydra for RSS feeds, auto-detect
repository URLs and clone instead of HTTP crawling.

- repo.py: detect_vcs(), clone_repo(), walk_files(), symlink_to_vault()
- Symlink storage: files in repo_vault/ symlink to content-addressed vault/
- Database: repo_uri, repo_path, commit_hash, vcs_type columns
- Auto-detect GitHub, GitLab, Bitbucket, sr.ht, hg.mozilla.org, etc.

database.py

@@ -110,6 +110,11 @@ class MediaSource(Base):
     searchable_text = Column(Text)
     crawl_job_id = Column(Integer, ForeignKey('crawl_jobs.id'))
     discovered_at = Column(Text, nullable=False)
+    # VCS repo metadata (for code files from git/hg/svn clones)
+    repo_uri = Column(Text)       # Clone URL
+    repo_path = Column(Text)      # File path within repo
+    commit_hash = Column(Text)    # Commit/revision when indexed
+    vcs_type = Column(Text)       # git, hg, svn, fossil
 
     media = relationship('Media', back_populates='sources')
     crawl_job = relationship('CrawlJob', back_populates='media_sources')
@@ -120,6 +125,7 @@ class MediaSource(Base):
         Index('idx_sources_job', 'crawl_job_id'),
         Index('idx_sources_media_uri', 'media_uri'),
         Index('idx_sources_page_uri', 'page_uri'),
+        Index('idx_sources_repo', 'repo_uri'),
     )
 
 
@@ -485,6 +491,12 @@ class Database:
         detail_content: str = "",
         searchable_text: str = "",
         score: int = SCORE_THUMBNAIL,
+        # VCS repo metadata
+        repo_uri: str = "",
+        repo_path: str = "",
+        commit_hash: str = "",
+        vcs_type: str = "",
+        keywords: list = None,
     ) -> None:
         """Create a new media record and add source context."""
         now = datetime.now(timezone.utc).isoformat()
@@ -521,7 +533,11 @@ class Database:
                 detail_content=detail_content,
                 searchable_text=searchable_text,
                 crawl_job_id=crawl_job_id,
-                discovered_at=now
+                discovered_at=now,
+                repo_uri=repo_uri or None,
+                repo_path=repo_path or None,
+                commit_hash=commit_hash or None,
+                vcs_type=vcs_type or None,
             ).on_conflict_do_nothing()
             await session.execute(source_stmt)
             await session.commit()